Cargando…

Automatic Verification of the Gradient Table in Diffusion-Weighted MRI Based on Fiber Continuity

In diffusion-weighted magnetic resonance imaging (dMRI), the coordinate systems where the image and the diffusion gradients are represented may be inconsistent, thus impacting the quality of subsequent fiber tracking and connectivity analysis. Empirical verification of the reconstructed fiber orient...

Descripción completa

Detalles Bibliográficos
Autor principal: Aganj, Iman
Formato: Online Artículo Texto
Lenguaje:English
Publicado: Nature Publishing Group UK 2018
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6224393/
https://www.ncbi.nlm.nih.gov/pubmed/30410013
http://dx.doi.org/10.1038/s41598-018-34940-4
Descripción
Sumario:In diffusion-weighted magnetic resonance imaging (dMRI), the coordinate systems where the image and the diffusion gradients are represented may be inconsistent, thus impacting the quality of subsequent fiber tracking and connectivity analysis. Empirical verification of the reconstructed fiber orientations and subsequent correction of the gradient table (by permutation and flipping), both time-consuming tasks, are therefore often necessary. To save manual labor in studies involving dMRI, we introduce a new automatic gradient-table verification approach, which we propose to include in the dMRI processing pipeline. To that end, we exploit the concept of fiber continuity – the assumption that, in the fibrous tissue (such as the brain white matter), fiber bundles vary smoothly along their own orientations. Our tractography-free method tries all possible permutation and flip configurations of the gradient table, and in each case, assesses the consistency of the reconstructed fiber orientations with fiber continuity. Our algorithm then suggests the correct gradient table by choosing the configuration with the most consistent fiber orientations. We validated our method in 185 experiments on human brain dMRI data form three public data sources. The proposed algorithm identified the correct permutation and flip configuration for the gradient table in all the experiments.